Optimization of time-domain header processing filters using genetic algorithms

We present genetic algorithms as optimization tools for the design of header recognition filters based on time-delay lines. The fitness parameter in the selection process is the discrimination of the target against all other bit patterns. For phase-shift-keyed headers, optimized filters exhibit improved discrimination compared with the matched filter. Time-domain filtering with no ambiguities is also achieved for specific amplitude-shift-keyed headers. Delay-line-based time-domain filters can be implemented in a planar lightwave circuit platform by simple phase and amplitude adjustments on the time delay taps.